SAPTARSHI BASU | EXTRA FIRST CLASS | September 15, 2018

Structural strength of ship is of paramount importance and a thorough

understanding the stresses on the ships structure is required along with the
possible modes of failure. Discuss the various modes of structural failure with
particular attention to: -

(a) Resistance of ship structure to bucking

(b) Strength of transverse structure
(c) Ultimate failure (due to Caldwell)

1) Overview
The history of maritime world is full of various accidents, many of them being due to
structural failure. It is therefore important for naval architects to keep in mind the various
loads that a ship could be subjected to, in her entire lifetime in the worst possible
scenarios, when designing the ship. During World War II, over 200 welded-steel ships broke
in half due to brittle fracture, caused by stresses created from the welding process,
temperature changes, and by the stress concentrations at the square corners of the
bulkheads.
Although several accidents occur due to negligence or other elements such as fire, failure in
ship structure is much more likely to occur due to a combination of fatigue and corrosion.
We have already studied about the ship hull design and know that the hull is the safe
casing which prevents water ingress apart from other support functions. So, let us study
about the various modes of failure leading to ship hull breach, along with the causes of
those failures. The wreck of RMS Titanic is arguably the most famous marine casualty of
modern times. On 14 April 1912 during her maiden voyage, RMS Titanic struck an iceberg
southeast of Newfoundland, Canada. She floated for approximately two hours, eventually
assuming an extreme trim by the bow and breaking in half. 705 people survived the loss of
RMS Titanic, so there was no shortage of eyewitnesses. Unfortunately, accounts differ
greatly, with some survivors claiming that the vessel plunged intact and some that her
stern rose out of the water and then broke off, both pieces ultimately sinking. There is no
debate that the primary cause of the demise of the Titanic was the collision with ice. It is
the contributing factors that caused this “unsinkable” ship to plunge after only two hours,
claiming 1,517 lives that will continue to be the fascination of engineers. Popular theories
hold that the watertight bulkheads were not continued high enough in the ship, allowing
flooding over the top; brittle fracture occurred due to poor quality steel and or low
temperatures; and that the stern rose out of the water to as much as 40 degrees of trim
prior to braking apart. SNAME’s Marine Forensic Panel has devoted substantial resources
to investigating the Titanic and has published several of the most scientific papers on the
topic.
Fatigue and fracture are important design criteria for ships to ensure a sufficiently high
safety level. Ship structures are prone to fatigue because of high cyclic loading caused
primarily by waves, but also because of changing cargo distributions and excited
vibrations. After describing these types of loading, the chapter outlines the factors affecting
fatigue strength in addition to stress ranges, which are material type, mean and residual
stress, fabrication factors like quality and weld imperfections, and finally the size and plate
thickness effect. Subsequently, fatigue strength assessment methods mainly applied to ship
structures are dealt with, i.e. the S‐N approaches such as nominal, structural hot‐spot, and
notch stress approaches in conjunction with Palmgren‐Miner rule. The approaches are
PAGE 1
 illustrated by an application example of a tanker in detail. The second part covers unstable
fracture. After describing the types of fracture and possible countermeasures, fracture
assessment methods are briefly outlined, explaining different crack driving parameters and
the so‐called failure assessment diagram (FAD) as an example for assessing the safety
against unstable fracture.
Buckling or structural instability is considered one of the main modes of failure of ship
structural elements. The stability phenomenon of ship structures is defined by the state of
equilibrium of structural members. The equilibrium of the designed structure is stable if
small imperfections and defects will cause correspondingly small deviations from the
idealized operating conditions. If small imperfections cause disproportionately large
deviations, the equilibrium is unstable. All idealized structure differs from the actual
structure due to small deviations, defects and imperfections. Despite the presence of these
deviations, the actual structure should operate and perform in a manner similar to its
corresponding idealized structure. Therefore, the size of structure (geometry and scantlings)
should be selected to ensure that stable equilibrium would occur under all kinds of
perturbations. Thus, the stability concept describes the relationship between the
perturbing causes and the resulting consequences. The equilibrium of the system is called
stable if the system returns to the initial equilibrium state after the removal of a small
disturbing force.
The development of reliability-based
design criteria for surface ship
structures needs to consider the
following three components:
(1) loads,
(2) structural strength, and
(3) methods of reliability analysis.
A methodology for reliability-based
design of ship structures is
provided in this document. The
methodology consists of the
following two approaches: (1) direct
reliability-based design, and (2) load
and resistance factor design (LRFD)
rules. According to this
methodology, loads can be linearly
or nonlinearly treated. Also, in assessing structural strength, linear or nonlinear analysis
can be used. The reliability assessment and reliability-based design can be performed at
several levels of a structural system, such as at the hull-girder, grillage, panel, plate and
detail levels. A rational treatment of uncertainty is suggested by considering all its types.
Also, failure definitions can have significant effects on the assessed reliability, or resulting
reliability-based designs

2) Introduction
Fatigue and fracture are significant failure modes of ships as history and recent service
experience have shown. Associated failures have to be avoided or at least controlled to
exclude costly repairs and catastrophic events. Fatigue and fracture became significant in
the middle of the twentieth century. Well known are brittle fractures of several standard
ships such as Liberty freighters and T2‐tankers built during the Second World War where
PAGE 2
 material with low fracture toughness was used. Moreover, the introduction of welding
associated with higher utilization of material increased the number of failures due to
fatigue and fracture. Particularly increased cyclic stresses at the notches of welds caused
several fatigue cracks in ship structures. Enlarged local loads acting at supertankers built
in the 1960s and 1970s exaggerated the problem of fatigue failures, resulting in thousands
of cracks in several ships after few years of service. In addition, navy vessels were affected,
particularly structures made of aluminum alloys. These failures called for consideration of
fatigue during design, which was introduced by major classification societies in the 1990s
after reoccurrence of fatigue failures in tankers built of higher‐tensile steels, which resulted
in higher stresses.
Today, knowledge of fatigue strength of materials and welded joints is so far established
that the design of fatigue‐resistant ship structures should be possible as long as local
stresses at all structural details can be predicted. This might create problems still today.
Regarding unstable fracture, the material toughness has been improved, and assessment
methods based on experience as well as fracture mechanics have been developed so that
associated failures are rare today. Nevertheless, this subject still needs attention in view of
potential failures which may result in catastrophic events. The following sections give an
overview about causes and influence factors of fatigue and unstable fracture, followed by
assessment methods related to ship structures, illustrated by a typical example for fatigue
assessment.

3) Modes of Failure of Ship Hulls

From the loading point of view, a ship is essentially an elastic beam floating on the water
surface and subjected to a range of fluctuating and steady loads. These loads will generate
bending moment and shear force which may acting on the ship.
Forces on the hull include:
a) Action of sea
b) Force acting on the heavy items composed of gravity forces and the dynamic forces due
to acceleration imparted by the ship's motion
c) Thrust due to main propulsion forces on the hull

Failure is said to occur when the structure can no longer carry out its intended function
Although most structural components are designed in such a manner that if one element
fails, it sheds it load on to another element which can withstand this additional load, hence
normally this does not lead to an immediate catastrophe but immediate remedial action is
certainly necessary to prevent further damage.
However, there is a phenomenon known as the "domino" effect, wherein the surrounding
structural elements fail in succession and the result can be loss of the ship, hence the
necessity of immediate action mentioned above.
Ship structural failures are nearly always nonlinear, either a geometric nonlinearity
(buckling or large deflections) or a material nonlinearity (yielding and plastic deformation).
For steel members, the three basic types of failure and their subdivisions are as follows.
1. Large local plasticity
2. Instability Fracture,
3. Direct (tensile rupture), fatigue, brittle, elastic and inelastic buckling

PAGE 3
 The multi modes of failure includes combined elastic buckling and inelastic buckling and
plasticity,

Failure limit states of ship structures

4) Categories of ship structural failures

Ship structural failures

can be categorized
 Hull girder failures
 Local structural failures
These types of failures
could be categorized as
cracks, fractures, buckling
and excessive
deformations. The modes
of failure are categorized
as fracture and non-
fracture types.
PAGE 4
 The frequency of buckling failure of the deck structure of general cargo ship is shown in Fig

Buckling failure of deck structural members

The structural failures

of brackets in general
cargo ships are
as shown in Fig.

5) Causes of Hull Failure

The main causes of ship structural failures are
- Design errors
- Material errors
- Fabrication residual stresses, distortions and errors,
- Operational, maintenance and repair errors

Listed below are three reasons as to why the ship's hull may fail.
2. Failure may occur due to the hull structure becoming distorted on being strained
beyond the yield point. This will lead to a permanent set and distortion of the structure.
3. Failure may also be caused due to cracking of the hull structure. This occurs when the
material can no longer sustain the load applied on it various sections. The loading may
exceed the ultimate strength of the material or more likely the failure is due to fatigue of
the material leading to a cracking and ultimate fracture. Even if the cracks in the hull
do not lead to complete fracture, they can prove disastrous as it can lead to leakage and
thus cause flooding and consequent sinking of the vessel,

PAGE 5
 4. Failure may also be due to the structural instability. Very large deflections can occur
under relatively lighter loads, and the structure behaves like a crippled strut. We will
learn about the various stresses occurring on the ship's hull in our next article.

6) Buckling mode of failure of ship structure (Instability)

Buckling or instability causes excessive lateral deflection. In a strut or columns, lateral
deflection occurs. In a plate panel, wrinkling occurs, in a cylinder under radial pressure,
corrugation of circumference occurs, in a plate stiffener combination, torsion tripping may
occur, etc. These modes of failure are characterized by a relatively rapid increase in
deflection for a small increase in load. Buckling of ship plates, stiffening members and
stiffened panels represent one of the main modes of ship structural failures. The main
causes of buckling failure of ship structural elements are shown in Fig

Causes and consequences of

buckling failure
Unacceptable deformations
A structural element having sufficient strength may not necessarily have sufficient
stiffness. Large deflections may therefore take place. Although these deflections, or
deformations, may not cause structural failure they may have adverse effects on structural
performance, particularly under compressive and dynamic loading. Deficient structural
stiffness may cause excessive deformations and amplitudes of vibration.

Excessive deformations

Impact of corrosion on structural strength and stiffness

• Reduction of local and hull girder scantlings
• Reduction of load carrying capacity
• Increase of hull girder and local stresses
• Reduction of hull girder and local flexural rigidity
• Reduction of buckling strength
• Reduction of fatigue strength, etc.
The effect of corrosion of the
deck plating on the deck section
modulus and ultimate strength
of the ship is shown in Fig

Effect of corrosion of
deck structure on
ultimate bending
moment PAGE 6
The effect of corrosion on the flexural rigidity of a panel of plating for different plate
thicknesses is shown in Fig

Effect of plating
wastage on
flexural rigidity

Variation of plate
buckling strength
with time
(Exponential Model)

The effect of corrosion on the flexural buckling of a plate panel is shown in Fig. for an
assumed exponential model of plate deterioration and Fig. for an assumed parabolic model
of plate deterioration. Fig. assumes that the variation of plate thickness with time is given
by:

Assuming an exponential model to represent the variation of plate thickness with time, the
following model could be used:

Where: -

1) to =original plate thickness before start of corrosion.

2) tt = plate thickness at time “t” after corrosion
3) a and b factors dependent on the rate of corrosion.

The effect of corrosion on the

magnitude of the section modulus
of frames and longitudinals having
different configurations are shown
in Fig

PAGE 7
 Typical examples of buckling failure of ship structure

1- Buckling of the web plating of the

transverses in the top wing tanks of a
bulk carrier is shown in Fig.
2- Buckling of the floor plating in the
double bottom of a general cargo ship
is shown in Fig

A typical plate buckling of a stiffened

panel, see Fig

Mode of buckling of bottom plating of

oil tankers

Plate buckling of a stiffened panel

7) Hull Girder and Primary Loads

Traditionally, longitudinal strength has been determined by balancing the ship on a static
wave. This approach has been widely accepted as an expedient means of simplifying a
time dependent dynamic situation into a simple static analysis. The ability to meet
operational requirements using a static balance method is implicitly based on the
historical success of the method. The standard wave height used by the U. S. Navy in this
procedure is 1.1√(LBP), where LBP is the length between perpendiculars in feet and 1.1 is
an empirical coefficient. The ship is balanced on a trough, resulting in a sagging design
condition and on a crest, resulting in the hogging design condition. Longitudinal bending
moments and shears are then determined by treating the ship as a free-free beam.

Typically, a ship is divided into 20 stations between the forward and aft perpendiculars.
Cross sectional beam properties and primary stresses are determined for each station. To
simplify design calculations a stress, envelop is assumed taking the design primary stress
limit value as constant throughout some portion of the midbody length dictated by
judgement. Fore and aft, the design primary stress tapers to zero. This calculated stress
must be below the design stress by a certain stress factor (margin) to account for future
growth in displacement. This stress factor varies from 0.5 Tsi to 1.0 Tsi depending on the
ship type. The calculated primary stress cannot exceed design stress values, otherwise
PAGE 8
 additional material must be added to lower hull girder stresses. The design primary stress
limit, which varies from 8.5 Tsi to 10.5 Tsi depending on material, are based on past
experience and are empirical in nature. Indirectly they provide a check on fatigue.

As an alternative to the static balance method, the development of criteria based on

probabilistic methods is desirable. Such a method offers a unified approach to structural
design limiting values for fatigue and maximum environmental loading, defines the
dynamic components of the seaway response for specific operational requirements, and
establishes probability of exceedance for a given design level. Probabilistic methods are
also desirable from the perspective of translating operational requirements, such as area of
operation and expected ship life into design loads and strength limits.

Fatigue prediction has become increasingly important due to extended ship lives and the
greater use of higher strength steels to accommodate increased payload. Both of these
trends have resulted in increasing primary stress levels, which in turn cause greater
fatigue damage. In a recent naval ship design the requirement for adequate fatigue life
translated to a maximum allowable stress range with corresponding structural details.
This stress range then defines the minimum hull girder section modulus required. The
maximum permissible stress range replaces the design primary stress limit as a fatigue
check on primary stress. The maximum permissible range is linked to the service life, the
expected construction details and the area of operation assumed for the ship.

8) Secondary Loads
The Navy has historically used a first principles approach in sizing structure, typical
operating secondary loads are combined with primary loads to check structure for yield,
buckling, ultimate strength and torsional stability. External hydrostatic loads are treated
as static and are determined from empirically based formulas. Typically, live loads are
historically based pressures, which are accentuated to include motion effects. Tank
pressures are based on actual ship parameters, such as the overflow height. Vehicle
reactions are covered in great detail; they are calculated using a static balance. The effects
of ship motion are included in the calculation of forces on the vehicle. The use of first
principles, while more labor intensive then typical Class rules, has provided a more
accurate determination of ship structural requirements, and allows for greater freedom and
versatility in developing scantlings. Ultimately design criteria which links hydrostatic hull
pressures to hull girder bending should be developed. This will allow for consideration of
phasing between primary and secondary loads. The probability based secondary loads
provides a means of assessing fatigue performance of transverse structure and
connections.

9) Fatigue Loading and Affected Structures This section describes at first the relevant load
effects causing fatigue. Subsequently, the areas in ship structures are reviewed which
require particular attention during fatigue design. Fatigue generally occurs in conjunction
with fluctuating (cyclic) loads and hence, stresses. Cyclic loads can be subdivided into three
categories according to their frequency and/or period:

a) Loads Due to Varying Loading Conditions The changes of cargo‐loading conditions and
associated drafts generally cause load fluctuations with rather low frequencies,
respectively, long periods, the latter varying between hours for ferries and weeks for

PAGE 9
 ships sailing over long distances, thus resulting in hundreds to thousands of load
cycles during the operational lifetime.
b) Loads in a Seaway Wave‐induced pressure and ship motions in the seaway generate
cyclic stresses with relatively low frequency that is usually in the order of 0.1 Hz, or
periods around 10 s. For an operational lifetime of 25 years, the total number can be
between 107 and 108 load cycles
c) Propulsion‐Induced Loads and Vibrations Higher‐frequent loads caused by engines and
propellers result in forced vibrations with high number of load cycles, typically 1010or
more. In addition, wave impacts (slamming) as well as small regular waves may excite
hull girder vibrations (whipping and springing) in the order of 1 Hz being superimposed
to the wave loads mentioned above.

In general, it can be stated that all structural members subjected to high cyclic stresses are
prone to fatigue. The fatigue loading may be due to:

1) waves (particularly in deck and side structures of the hull girder)

2) propulsion forces (mainly close to main engine and propeller)
3) steering forces (mainly in rudder including stock and around bearings)
4) changing loading conditions (particularly in transverse structural members)
5) cargo handling (in cranes, ramps, etc. including foundations)

The fatigue criticality is governed by the severity of the loading, characterized by the shape
of the stress range spectrum and the number of load cycles, as well as the amount of the
stress concentration and mean stress. The majority of past fatigue failures can be
attributed to wave‐induced loads, particularly in structural members at the ship's sides and
longitudinal members of the hull girder particularly in deck. The fatigue failures in tanker
structures mentioned in Section 1belonged to the first ones.

Plastic collapse under

sagging/hogging moments.
Inelastic buckling of the deck
and bottom structure of an
oil tanker is shown in Fig.

10) Role of Classification Societies in controlling ship structure failures Classification

societies play a major role in ensuring adequate safety of ship hull girder and local
structural details, see Fig.
Most of ship structural failures result from:

- Design errors
- Material errors
- Fabrication errors
- Excessive corrosion (lack of proper maintenance).
PAGE 10
Ship designers should also pay much attention to the design of ship structural details, as
poor design of these structural details is responsible for most of the local structural
failures. Control of buckling of a plate panel subjected to compressive loading could be
achieved by fitting a stiffener as shown in

Role of Classification Societies in controlling ship structure failures Control of ship structural failures

Improving plate
buckling by adding a
stiffener

Control of buckling
failure of plate
panels under shear
loading

Plate buckling failure of the transverse webs of top wing tanks of bulk carriers could be
repaired by adding stiffeners

Buckling of the web frames of

the top wing tanks and
correction measures

We have had a lot of marine

accidents that involved failure of the
hull structures. Whether it was a
crack in the midship region, or a total split-off of the hull girder, or failures due to
propagation of cracks, the crux of the matter boils down to a handful of causes that are of
great concern to ship designers and operators. Mostly, crack propagation takes place due to
fatigue. When the hull girder of a ship is designed, the designers analyze the structure as a
beam. the buoyancy on the hull is never predictable given the fact that sea surface being
PAGE 11
characterized by waves, the buoyancy on the ship is always varying periodically along the
ship’s length. Also, ships are not always in the same cargo loading conditions. Where they
may ply one voyage in a fully loaded condition, in the return voyage the ship may not have
cargo but be induced to ballast loading condition.

Figure 1 shows schematically the stress history in a hull girder due to the loads mentioned
under 1 and 2. The history is typical for ships with significant changes between voyages in
ballast and fully loaded, for example, in tankers and bulk carriers. Figure 2 shows the
measured stress history over 80 s with low‐frequent stress fluctuation due to waves and
higher‐frequent one due to bow‐flare slamming in a container ship. The first one has been
derived by low‐pass filtering. The latter causes hull girder vibration with a frequency of 0.5
Hz

Figure 1 Stress history for superimposed Figure 2 Stress history from on‐board
Stillwater and wave‐induced loads measurement.
(schematically).

The relevant parameters for the fatigue damage process, that is, stress range Δσ of the
individual cycles and mean stress σ m, can be derived from a measured or simulated history
by different cycle‐counting methods. The rainflow method (Matsuishi and Endo, 1968) is
preferred as it considers the related elastic–plastic material behavior causing the fatigue
damage. The evaluation of long‐term measurements has resulted in typical frequency
distributions of stress ranges due to the seaway which are usually represented as
cumulative distributions of stress ranges (also designated as stress range spectra). These
can frequently be approximated by two‐parameter Weibull distributions (Almar‐Næss, 1985)
with the following probability Q for exceedance of a stress range Δσ:

where n is the number of stress cycles that exceeds Δσ, Δσ max the maximum stress range for
a total of n max cycles, n max the total number of stress cycles, and h the Weibull shape
parameter. Several long‐term measurements of wave‐induced stresses have shown that the
shape parameter is close to unity which results in the so‐called straight‐line spectrum in a
linear‐logarithmic representation. Such standardized distributions are usually assumed
during fatigue design, where the parameters, that is, the characteristic stress range and
number of stress cycles, are either directly computed or based on the assumptions given in
the rules of the classification societies.

PAGE 12
11) Structural Areas Prone to Fatigue
In general, it can be stated that all structural members subjected to high cyclic stresses are
prone to fatigue. The fatigue loading may be due to:

 waves (particularly in deck and side structures of the hull girder)

 propulsion forces (mainly close to main engine and propeller)
 steering forces (mainly in rudder including stock and around bearings)
 changing loading conditions (particularly in transverse structural members)
 cargo handling (in cranes, ramps, etc. including foundations)

The fatigue criticality is governed by the severity of the loading, characterized by the shape
of the stress range spectrum and the number of load cycles, as well as the amount of the
stress concentration and mean stress.

The majority of past fatigue failures can be attributed to wave‐induced loads, particularly in
structural members at the ship's sides and longitudinal members of the hull girder
particularly in deck. The fatigue failures in tanker structures mentioned in
Section 1belonged to the first ones. Figure 3 shows the affected area below the loaded
waterline (LWL), where high pressure fluctuations between wave crest and trough caused
fatigue cracks at the intersection between side longitudinals and transverse ring structures.
Particularly, the welded connection between web stiffener and side longitudinal was critical,
and also the cut‐out in the transverse web. In addition to any welded attachment on the
side longitudinal, the transverse structures are prone to fatigue, especially the knuckles of
the flanges including the transition to hopper and wing tanks and the toes of large
brackets.

Figure 3 Fatigue
crack at a side
shell longitudinal
of a large tanker
in the 1990s.
(Reproduced with
permission from
Yoneya et
al., 1993. ©
EMAS Publishing,
1993.)

The longitudinal members particularly in deck are mainly stressed by vertical hull girder
bending moments, and also by horizontal hull girder bending and torsion, for example, in
container ships. Prone to fatigue are again all attachments and discontinuities of
longitudinal members. This includes hatch corners, terminations of hatch coamings and
superstructures as well as access openings, windows, and other cut‐outs. In addition,
outfitting elements connected to longitudinal members as well as hatch cover supports in
container ships require special attention, the latter being subjected to friction forces during
deformations of the ship. Up to now, design of fatigue‐resistant ship structures requires
PAGE 13
 much experience and knowledge about the distribution of cyclic stresses as well as stress
concentrations.

12) Factors Affecting Fatigue Strength

The fatigue strength of a structure is mainly influenced by the number and amplitude or
range of stress cycles which result from the cycle counting of given stress histories or from
assumption for a stress spectrum. The stress range may include already notch effects. The
relationship between endurable stress ranges and number of stress cycles is given by the
Woehler S–N curve, see Figure 4.

Figure 4
Basic elements of the S–
N approach.
Reference is also made
to the basic literature
on fatigue, for example,
Almar‐Næss 1985,
Gurney 1979,
Radaj 1990, and
Maddox 1991.

Other factors affect the fatigue strength to a lesser degree, among them the following are
discussed in the subsequent sections:

(a) material
(b) mean stress and stress ratio
(c) residual stress
(d) quality and imperfections
(e) size and plate thickness effect

Ship structures are usually built to a defined set of details, which are documented by
classification societies, owners or the builders. Details include stringer and frame
intersections, bulkhead and stiffener connections (watertight and non-watertight),
penetrations, cutouts etc. The increasing importance of fatigue strength for maritime
structures has resulted in corresponding research activities and recommendations for
design, fabrication and operation of maritime structures to counter the increased risks
described above. These worldwide research efforts have produced an amount of
publications that it is impossible to give a comprehensive overview. The best starting point
for a less extensive literature surveys remains the proceedings of the ISSC (International
Ship Structure Conference).
In reality, fatigue problems appear after a certain number of load cycles, often only after
months or even years of operation. The feedback for design is then often slow. Experiments
allow accelerating the time scale in applying realistic loads at a much higher frequency and
PAGE 14
observing then fatigue problems (after the same number of load cycles) in much shorter
time, namely hours or days. Over several decades, test data were accumulated for many
typical ship structures and these data have been used by classification societies (along with
feedback from fatigue damages on actual ships) to compile catalogues of stress increasing
factors. These catalogues allow a simple, pragmatic approach to structural design. The
structural designer can compute macroscopic nominal stress using long established and
widely available standard tools for structural analysis. The catalogues then give a
correction for the influence of a discontinuity (like a weld or a corner) allowing to transform
fatigue strength into a changed upper limit for the static stresses. The approach is
pragmatic, but not applicable to structures which are not (yet) found in the catalogues.
This poses problems each time new structural details or new materials are used.
Design of ships is an interactive process, where major decisions are made in an early
design stage covering for instance main dimensions and general arrangements. Information
of structural details, which are basic requirements for reliable fatigue assessment, is
available in the following design stages. This is a significant obstacle for the early design
stage, because the decisions done in this stage have a strong influence on the fatigue life of
the hull girder. Structural modifications done after the early design stage are usually
limited and expensive for production. At present, there is no common approach for fatigue
assessment which can fulfill requirements of the early fatigue design.
Traditional dimensioning of ship structures followed simple semi-empirical formulae giving
directly the thickness of plates of stiffeners or a required section modulus. Fatigue
strength, however, requires a more detailed stress analysis. Complex ship structures are
increasingly analyzed using 3D finite elements models. These models allow a realistic
distribution of loads and capture the interaction between the main structures. Usually the
whole ship hull with its main structures is modeled using plate elements. Secondary
structures like stiffeners are modeled simplified using truss elements. The analysis gives
global nominal stress distributions for coarse grids.
For fine grids, effects of effective width and geometry of the structure are also captured.
Local finite-element models serve to determine the stress increase due to geometry of the
structure. Usually plain-strain plate elements suffice to determine the notch stress at plate
edges of holes. For hot-spot stress at weld toes and plate structures, all plate elements of
volume elements are employed. Volume elements require more effort, but consider the
stiffness and load distributing effect of the weld better. The definition of the hot-spot stress
requires the evaluation of the linear stress component over the plate thickness. This is
automatically given for plate elements. For volume elements, an elegant solution is
arranging only one element over the plate thickness. Then intermediate nodes are
necessary at the element edges to capture the bending properly.
Using only 2 integration points over the thickness yields directly the linear stress
component. This can then be extrapolated to the plate edge to give the hotspot stress. The
loads for the local finite-element models come either from prescribed external stresses or
deformations. These are taken either from a defined initial state or from a global analysis.
Based on a scientific and engineering approach, fatigue-critical structural details and
important characteristics of the ships are assessed. The scientific approach is based on
damage statistics of fatigue failures in ship structures, and aims to identify critical
structural details and corresponding loading modes.

PAGE 15
In the European Project IMPROVE the review is concentrated especially on tankers, and is
extended to cover Ropax and LNG ships based on an engineering approach. The
engineering approach is based on pre-existing know-how and knowledge. The study is
focused on special features of different ship types which are further developed in the
Improve project. The main results of this study are the identification of generic and ship-
type-depended features in fatigue assessment. This is the basis for the development of the
fatigue approach for early design stage.
The important calculus is focused on the hotspot and notch stress level, and it is based on
the results of extensive FE-analysis of typical structural detail, performed in SDG, as
sensitivity analysis. In general, there are several approaches for fatigue assessment. The
commonly used methods can be divided into different groups according to the applied
strength parameters and corresponding response analysis. In this figure, starting from left
to right side, the methods are divided to the global and local approaches. The global
approaches, such as nominal stress approach, are based on main dimension of the
structure. These approaches are easy to apply, but their practical applications for complex
ship structures are limited. Therefore, the use of more advanced approaches considering
local parameters are preferred and applied (ISSC 2003, IACS 2008). These methods in ship
design are structural hot-spot stress and notch stress approaches. The fracture mechanics
approaches with J-integral or stress intensity factors are not commonly used in the design
stage due to the extremely time-consuming structural analysis.
Weakness of the structural hot-spot stress and notch stress approaches is that they require
structural analysis in a detailed level. This can be obtained by applying the finite element
method, which is however time consuming and, thus, not suitable for the early design
stage. At present there is no common and simplified approach for fatigue assessment which
can fulfill the requirements of early fatigue design. Estimating the production cost is a
fundamental part of ship design. Traditionally, shipyards employ empirical methods to
estimate the cost of a new ship, as ships typically are one-of a-kind products and orders
are won based on early bids, i.e. naval architects must estimate costs a priori and within
relatively short time (order of several days). Most of these traditional cost estimate
approaches are related to the ship weight, which in turn is estimated based on ship type
and main dimensions as well as installed power and equipment and outfit.
(a) Material: - The fatigue behavior is influenced to a large extent by the type of
material. For instance, ship structures made of aluminum alloy have a fatigue
strength being about 2.5 times lesser than those made of steel. Within one material
type, the material strength has some influence on the fatigue strength of relatively
mild notches, whereas the influence has found to be negligible for structures with
relatively high notch effects, for example, for welds. A direct consequence is that
special attention has to be paid to the detail design of ship structures made of higher
strength materials where the permissible stresses and hence cyclic stresses are
higher. Reduction of notch effects and post-weld improvements are recommended
here.
(b) Mean stress and stress ratio: - High mean stress normally reduces
the fatigue strength. For loading with constant stress ranges, the
mean stress is characterized by the stress ratio R:
where σ l is the lower and σ u the upper stress of the stress cycle
considered. The mean‐stress sensitivity is usually described by the difference
between the fatigue strength in the high cycle domain for R = −1 (pure alternating
stress) and that for R = 0 (tensile pulsating stress), divided by the second one.
PAGE 16
 Different mean‐stress sensitivities have been found for different materials with a clear
trend toward increased sensitivities for higher strength materials of the same type.
Not only crack initiation is affected by mean stress, but particularly crack
propagation due to crack closure during the compressive part of the load cycle.
(c) Residual Stress: - Residual stresses and distortions are induced by the fabrication
process, in particular by welding due to the high heat input. These residual stresses
are superimposed to the stresses due to external loading so that they act in a similar
way as mean stresses. High-tensile residual stress may be present at weld toes due to
the cooling process after welding which may even reach the yield stress. In this case,
the yield limit is exceeded in the first load cycle, causing partial relaxation of the
residual stress such that the superimposed stress just reaches the yield stress in the
following load cycles again. Therefore, large stress cycles in welded joints generally
reach the yield stress if notable tensile residual stresses are present. This means that
the mean stress has no remarkable effect on the fatigue strength in structures with
high residual stresses
(d) Quality and Imperfections: - Particularly, welded structures are never perfect and
contain deviations in shape including misalignments as well as irregularities such as
undercuts, slag inclusions, lack of fusion, and so on. As long as these are within
permissible limits, they are called imperfections, otherwise defects requiring corrective
measures. The imperfections mentioned are—in addition to material effects—mainly
responsible for the relatively large scatter in fatigue life observed in tests of small‐
scale specimens with the same geometry and type of loading. Particularly, axial and
angular misalignments affect the fatigue strength of axially loaded structures due to
the formation of secondary bending stresses, which can be regarded as structural
stress if added to the nominal axial stress so that they can explicitly be considered
during fatigue assessment.
(e) Size and Plate Thickness Effect: - The last parameter to be mentioned particularly
for welded structures is the so‐called size effect which reduces the fatigue strength, if
the size of the structural component becomes larger. In addition to the statistical
effect, meaning that the probability of weak points increases with the size of the
structure, the plate thickness effect is an important design parameter.
The decrease in fatigue strength of welds for increased plate thickness can be
explained by two effects:
 As the weld toe radius remains more or less constant, the ratio between toe radius
and plate thickness decreases so that the notch stress increases.
 The residual stress field is often characterized by high tensile stress at weld toes
and—for equilibrium reasons—compressive stresses in between. In thicker plates, the
tensile zone is spread over a larger depth of the plate so that the propagation of a
crack initiating at the weld toe will be much faster.
 The plate thickness effect is usually considered only above a reference
thickness t0 with the following reduction factor ft on fatigue strength proposed by
Gurney 1979

Where:

The exponent n varies between 0.1 for longitudinal attachments welded to plate
edges, 0.2 for butt‐welded joints, and 0.3 for fillet‐welded joints (Hobbacher, 2009).
The reference thickness t 0 defined in rules and recommendations ranges from 22 to
25 mm for plate‐type structures. Below this, no positive thickness effect is assumed

PAGE 17
 which can be justified by the counteracting effect of misalignments becoming more
severe in thinner structures.

Design for fatigue leads to much more elaborate structural design and assembly
procedures. In ISSC 2009 Committee III.2 Report, fatigue design methods for ship and
offshore structures are discussed. For ships, rule-based-methods for fatigue evaluation are
proposed by classification societies, and a comparison of the different fatigue methodologies
provided by BV, DNV, GL, KR, LR and NK is summarized. The IACS Common structural
rules (CSR) for Oil Tankers and Bulk Carriers Hull structure, published in January 2006,
have been effective for new-builds since the 1st of April 2006 and it has been evaluated
extensively.
International Institute of Welding (IIW) provides fatigue recommendations of welded
components and structures, including the effect of weld imperfections with respect to
fatigue. The codes cover all current methods of verification, as e.g. component testing,
nominal stress, structural stress and notch stress methods, as well as fracture mechanics
assessment procedures. The safety philosophy covers the different strategies, which are
used in various fields of application and gives a specified choice for the designer. The main
areas of update are the structural hot-spot stress concept, which now allows for an
economic and coarser meshing of finite element analysis, the extension of the effective
notch stress concept to welded aluminum structures and numerical assessment of post
weld treatments for improving the fatigue properties.
Hobbacher presents the update of the recommendations for polishing welds to reduce
notch stresses. The structural weight is virtually unaffected, but the construction costs are
naturally significantly higher. Thus, a cost estimate following the traditional approach with
traditional empirical coefficients based on yesterday’s practice underestimates costs. New
approaches to cost estimates need to be developed to reflect modern design-for fatigue,
design-for-production approaches in ship building. Such approaches are pursued
worldwide at a few places and also subject to a research and development in Project
IMPROVE.
The development of the approach for fatigue assessment requires a balanced approach for
load, response and strength with sufficient accuracy. However, the approach should
overcome the challenge of limited information of structural details in the early design stage.
Simplification is also needed to obtain a generic approach, which is applicable for
optimization purposes and can be linked to the existing design tools. The existing design
tools have suitable databases of general geometry of structure and load specifications.
These design tools can be also applied to analyze primary and secondary stresses of hull
girder, and they should be exploited to fatigue analysis. Therefore, this study is focused
especially on the analysis of local stresses.
The main challenges are the transformation of the response from existing design tools and
the calculation of fatigue effective stresses in critical structural details with sophisticated
assumptions. This requires pre-defined structural details, which are generic, but however
define the fatigue strength of the hull girder.
Fatigue failures are usually detected in inspections of classification societies. While fatigue
problems are frequent, catastrophes due to fatigue are quite rare, because micro-cracks
take usually a long time (sometimes months to years) before progressing to structure
failure. They are then typically detected and rectified at an early stage. The inspection

PAGE 18
 requires surveyors (representatives of the classification society inspecting onboard) to look
at the structures up close. This poses problems in narrow spaces.
Some characteristics affecting the fatigue strength of the hull girder are strongly depended
on the ship type. The use of the ship is defining the main dimensions of the ship, shape of
the hull girder, geometry of the main frame and the steel arrangement. Sailing routes i.e.
weather condition and service speed can be different between different ships. These
differences affect mainly wave and cargo induced fatigue loading and response of hull girder
in nominal stress level. Structural details and connections are quite similar between
different ship types. However, to obtain optimum space for cargo transporting, special
structural elements such as pillars have been applied in some ship types.
13) Fatigue‐Strength Assessment

The fatigue strength of structures is usually assessed by the following approaches:

 S–N approach, using appropriate Woehler S–N curves either directly in the case of
constant amplitude loading or in conjunction with a damage accumulation
hypothesis in the case of variable amplitude loading.
 Crack propagation approach, normally using the Paris–Erdogan law for the crack
propagation phase up to a specified failure criterion. The crack initiation phase is
either neglected if an initial crack is assumed or if an existing crack is present.

In ship structural design, the S–N approach is applied in most cases, whereas the crack
propagation approach is used mainly for fitness‐for‐purpose assessment in case of an
existing crack found during new‐building or in‐service survey. For this reason, the S–
N approach is presented in the following. After describing the Palmgren–Miner rule for
damage accumulation, different variants of the approach are dealt with differing in the type
of stress used.

In total there are four approaches to fatigue design of the ships: -

1) Palmgren–Miner Rule for Damage Accumulation

2) Nominal Stress Approach.
3) Structural Hot‐Spot Stress Approach
4) Notch‐Stress and ‐Strain Approaches

The aim of the sensitivity analysis is to study the effect of the structural scantlings on the
structural hot spot stress factor. The analysis is focused on the end of slopping plate
marked by the red circles in figure 1.
The modelled structure is part of bottom and
side structure of LNG carrier. The analysis is
focused on the ends of tank top and slopping
plates marked by the red circles in figure 1.
The model includes, in the longitudinal
direction, between two web frames, and in the
transverse direction, part of the bottom and
side structure. In the sensitivity analysis the
scantlings such as thickness of tank top,
Additional panel stiffening of the
PAGE 19
ends of side girders in oil tankers.
 bottom and slopping
plates are varied.
The topology of the
structure is fixed to
simplify the
analysis.
Fatigue cracks tend
to occur in highly
stressed regions
with remarkable
change in geometry,
such as connection
between
longitudinal and
transverse
structures. The
highest occurrence
of fatigue cracks
was mainly observed amidships, where the maximum bending moment occurs. The
geometry of structural details has a huge influence on the fatigue strength. Three generic
structural elements can be identified that can cover hull girder structures: stiffened plates,
girders and pillars. However, it is important to notice that the loading mode of the
structural elements is different depending on its location within the hull girder and ship
type. The number of different structural details is relatively large causing challenges to
obtain the generic approach for the early design stage. Based on the results from the review
of fatigue damage statistics, it can be concluded that the end of longitudinal stiffeners,
particularly beam brackets and cut-outs are the most critical details. Important are also
the ends of slopping plates, which are fatigue-critical in the case of LNG ships. Several
different fatigue-critical details lead to the conclusion that some sophisticated grouping of
structural details will be required in the fatigue approach for the early design stage.
Fatigue for maritime structures has far more implications and aspects than ‘just’ the
experimental and numerical aspects of mechanical investigations. Ultimately, engineering
research is expected to yield better products and better procedures to support these
products (from design to operation). A wider view shows many facets that need to be
addressed. In my personal view, we will only succeed if we address the topic with a wider
view, without reducing the importance of focused or even fundamental research.
1) Palmgren–Miner Rule for Damage Accumulation

Basic elements of the S–N approach are illustrated in Figure 4, showing, from left to right

 stress spectrum, with maximum. stress range Δσ max and total number of load
cycles n max, for example, the Weibull distribution; please note that n on the horizontal
axis is the cumulative number of load cycles here; the spectrum is subdivided
into i blocks for numerical analysis, each containing ni cycles,
 S–N curve describing the fatigue life for constant amplitude loading, being usually
expressed above the knee point by

PAGE 20
 with the number of endured load cycles N, the stress range Δσ, the slope
exponent k and a constant K; beyond the knee point, the slope exponent is modified
according to Haibach 1969, that is to (2 m − 1), to account for variable amplitudes
below the knee point,
 life curve showing the number of endured stress cycles N for the same stress
spectrum with different maximum stress range Δσ max.
As mentioned, the life curve for variable amplitude loading is obtained from a damage
accumulation hypothesis, usually the linear Palmgren–Miner rule, giving the damage
sum D from partial damages at different stress levels i:

where ni is the number of stress cycles in block i and Ni the number of endured stress
cycles at level i according to the S–N curve, see Figure 4. Failure occurs when the limit
damage sum DL is reached, being usually defined between 0.5 and 1.0 (Hobbacher, 2009).
Alternatives to the summation of partial damages are closed‐form expressions (Almar‐
Næss, 1985) or tabulated values (Fricke, Petershagen, and Paetzold, 1997 and 1998) for
standard stress spectra following the Weibull distribution.
For design purposes, a conservative design S–N curve is usually applied. Typically, a
survival probability of 97.7% is suggested corresponding to a line lying two standard
deviations below the mean curve assuming Gaussian distribution of lives in logarithmic
scale. Unfavorable mean and residual stresses are usually assumed.

2) Nominal Stress Approach

The nominal stress σ n is defined as the stress being undisturbed by notches like cut‐outs
or welds. In a more general sense, it can be defined by integral load parameters and
sectional properties which allow its determination also in a section with locally increased
stresses, for example, by dividing the force by cross‐sectional area or the bending moment
by section modulus. However, macro-geometrical stress raisers which are not typical for the
detail, for example, due to effective breadth of wide or curved flanges, have to be included
in the nominal stress.

The nominal stress at welded joints refers to the section affected by the potential fatigue
crack, that is, the plate in front of the weld if the weld toe is critical, or the weld throat if
crack initiation is expected at the root of a nonpenetrating weld.

Basis for the nominal stress approach are S–N curves giving the fatigue strength of same or
similar details subjected to a well‐defined nominal stress. For this purpose, results of
existing fatigue tests were gathered and reevaluated with regard to the fatigue strength of
similar details in order to obtain design S–N curves for typical structural details.
Figure 5shows such S–N curves and associated welded steel details according to the
recommendations of the International Institute of Welding IIW (Hobbacher, 2009), being
characterized by the fatigue strength at two million cycles, the so‐called FAT class in
newton per square millimeter, and a slope exponent k = 3 above and k = 5 below the knee
point at 10 million cycles for variable amplitude loading. The curves are limited by the FAT
160 curve for parent material (steel) with k = 5.

PAGE 21
 Figure 5
(a) Selection of
design S–N curves
for nominal stress
approach and

(b) associated
structural details
according to
Hobbacher 2009

It should be noted that typical imperfections such as misalignment are already included in
the fatigue classes as these were present in the underlying fatigue tests. Similar design
curves and detail catalogs can be found in various codes and guidelines. Mostly applied in
the design of ships and offshore structures are those proposed by British Standard
(Maddox, 1991) and IIW (Hobbacher, 2009).

3) Structural Hot‐Spot Stress Approach

In contrast to the nominal stress, the structural hot‐spot stress σhs contains effects of the
structural configuration, such as the stress increase due to the longitudinal stiffener on a
plate, as shown in Figure 6. However, the local stress increase due to the weld toe is
excluded as indicated by the dashed line in the figure. Sometimes, the terms hot‐spot
stress and geometric stress are used, the latter because the structural geometry is
considered.

Distribution of longitudinal stress in

Figure 6
front of a longitudinal stiffener on an
axially loaded plate.
Different proposals exist for the
determination of the structural stress
at the weld toe. The most common way
is the extrapolation of the measured or
computed surface stress to the weld
toe. Generally, it is assumed that the
influence of the weld toe notch
vanishes in a distance of about
0.3t from the weld toe (Niemi, Fricke,
and Maddox, 2006), where t is the
adjoining plate thickness. Therefore,
stress values at reference points outside this area are used for extrapolation, for example,
at distances of 0.4t and 1.0t from the weld toe according to Hobbacher 2009 and Niemi,
Fricke, and Maddox 2006 or 0.5t and 1.5t according to several recommendations by
PAGE 22
classification societies. In both cases, the stress is linearly extrapolated. Recommendations
for finite element modeling and evaluation are given among others by Niemi, Fricke, and
Maddox 2006.

The structural stress can also be regarded as the sum of axial and bending stress in a plate
or shell, that is, the stress linearized at the hot‐spot in the through‐thickness direction as
also indicated in Figure 6. This clearly defined structural stress for plate‐type structures
can only be determined numerically. It coincides with the stress obtained by surface
extrapolation; however, slight deviations are possible depending on the extrapolation
procedure.

For attachments at plate edges, it is no more reasonable to use the plate thickness for
determining the extrapolation points nor to perform the through‐thickness linearization.
Instead, fixed points, 4, 8, and 12 mm away from the weld toe are recommended for stress
extrapolation along the plate edge (Niemi, Fricke, and Maddox, 2006; Hobbacher, 2009).

As the effects of the structural geometry, which are contained in the fatigue classes of the
nominal stress approach, are considered in the structural stress, a single S–N curve is
sufficient for each type of weld and material. Frequently, two S–N curves are defined, for
example, FAT 100 for butt welds and non-load‐carrying fillet welds and FAT 90 for partial‐
and full‐load‐carrying fillet welds at steel (Hobbacher, 2009; Niemi, Fricke, and
Maddox, 2006). The latter create a higher stress concentration being not visible in the
structural stress. It should be noted that pronounced effects of misalignment on the
structural stress are not contained in the fatigue classes, and have therefore to be
considered in the stress, particularly for butt joints and cruciform joints as well as for
transverse attachments because of possible axial and angular misalignment
(Hobbacher, 2009). In addition, plate thickness effects have to be considered. Variants of
the structural stress approach exist to consider further effects such as the stress gradient
in thickness directions or the plate thickness effect. Details are given by Radaj et al. 2006.

Calculated fatigue lives, calibrated with the relevant fatigue damage data, may give the
basis for the structural design (steel selection, scantlings and local details). Furthermore,
they can form the basis for efficient inspection programs during fabrication and throughout
the life of the structure. To ensure that the structure will fulfill its intended function,
fatigue assessment, supported where appropriate by a detailed fatigue analysis, should be
carried out for each individual type of structural detail which is subjected to extensive
dynamic loading. It should be noted that every welded joint and attachment or other form
of stress concentration is potentially a source of fatigue cracking and should be individually
considered.
According to [6], the structural hot spot stress σhs is determined using the reference points
and extrapolation equations for different refinement mess. For coarse mesh with higher-
order elements, having lengths equal to plate thickness at the hot spot, evaluation of
stresses at mid-side points or surface centers, respectively, i.e. at two reference points 0.5t
and 1.5t, and linear extrapolation is done with the formula

For the structure analyzed in the paper, according to the upper formula, stress variations
in the hot spots along the paths were obtained according to the linear variation (linear
extrapolation) in figure 6.

PAGE 23
The structural hot-spot stress is frequently determined by extrapolation from the reference
points mentioned before, particularly at points showing an additional stress singularity
such as stiffener ends. The structural or geometric stress σhs at the hot spot includes all
stress raising effects of a structural detail excluding all stress concentrations due to the
local weld profile itself. So, the non-linear peak stress σnlp caused by the local notch, i.e. the
weld toe, is excluded from the structural stress. The structural stress is dependent on the
global dimensional and loading parameters of the component in the vicinity of the joint. It
is determined on the surface at the hot spot of the component which is to be assessed.
Structural hot spot stresses σhs are generally defined at plate, shell and tubular structures.
The structural hot spot stress approach is recommended for welded joints where there is no
clearly defined nominal stress due to complicated geometric effects, and where the
structural discontinuity is not comparable to a classified structural detail.
The following considerations focus on extrapolation procedures of the surface stress, which
are nearly the same in measurement and calculation. Firstly, the stresses at the reference
points, i.e. extrapolation points, have to be determined, secondly the structural hot spot
stress has to be determined by extrapolation to the hot spot point. The structural hot spot
stress may be determined using two or three stress or strain values at particular reference
PAGE 24
points apart from the hot spot in direction of stress. This is practically the case of points
placed at the distances of 0.5 t and 1.5 t from the hot spot point, where t is plate thickness.
The structural hot spot stress at the hot spot is then obtained by extrapolation, according
to the equation (1).
For the LNG structural joint analyzed in the work, the stress concentration factor Khs has
values placed in the range: 1.5 < Khs < 4.0. This means that during the static or quasistatic
calculus of the LNG structure, for a certain load case, the values obtained for stress in the
hot spot point, for a coarse mesh will be multiplied by Khs to obtain the structural hot spot
stress σhs (according to equation 2). Identification of the critical points (hot spots) can be
made by:
a) measuring several different points,
b) analyzing the results of a prior FEM analysis,
c) experience of existing components, which failed

4) Notch‐Stress and ‐Strain Approaches

The elastic notch stress in a rounded corner or in a weld toe, which can numerically be
determined using linear‐elastic material behavior, considers the local geometry. The
increased stress in rounded corners, for example, at plate edges (cut‐outs, hatch corners,
etc.) can be assessed with respect to fatigue by utilizing the S–N curves of test specimens
with plate edges. Depending on the quality, which is determined by the cutting process and
possible post treatment like grinding, FAT classes between FAT 100 and 160 apply for steel
(Hobbacher, 2009).

The fatigue behavior of sharp notches is different as in

addition to the high, localized stress peak, the material
structure also plays a role preventing excessive local
yielding and supporting the notch root. Different
approaches exist to consider the related microstructural Figure 7 fictitious
support effects (Radaj et al., 2006). The approach by notch rounding of weld
Radaj 1990 has gained practical significance with toes and weld roots
PAGE 25
respect to welded joints. He introduced a correspondingly increased fictitious notch radius
of r ref = 1 mm for sharp weld toes and weld roots, assuming an actual radius of zero in the
worst case. Figure 7 shows the fictitious notch rounding of weld toes and roots for a
cruciform joint and a butt joint. At non-fused root faces, a so‐called keyhole notch is
arranged, placing the vertex point of the circle at the end of the slit, that is, the location of
the weld root. Usually the weld shape is idealized; however, the actual shape can also be
modeled.

Detailed aspects of finite element modeling and analysis are discussed by Fricke 2012. FAT
225 has been proposed for the design S–N curve of steel structures in this effective notch‐
stress approach (Hobbacher, 2009). It has to be emphasized that notch stresses may exceed
far the yield stress. A problem can occur with mild notches as the approach presupposes
naturally formed, as‐welded toes and roots. Therefore, an effective notch stress of at least
1.6 times the structural stress should be assumed. As in the structural hot‐spot stress
approach, pronounced misalignment effects have to be considered in the stress, however,
the plate thickness effect is naturally included.

Some recommendations of ship classification societies contain a very simple notch‐stress

approach by applying notch factors on the structural stress and using accordingly
increased design S–N curves, which principally is a factorized structural stress approach.

A more refined assessment may be offered by the notch‐strain approach which considers
elastic–plastic effects (Radaj et al., 2006). These may be relevant particularly in the low‐
cycle fatigue domain. The approach requires the so‐called cyclic material law of the affected
material. In case of welds, the parent metal, heat‐affected zone, and weld metal may show
different behavior which complicates the analysis.

The local stress and strain are normally determined by nonlinear finite element analyses or
approximation formulae (e.g., Neuber's rule). The fatigue strength is described by strain S –
N curves. Contrary to the other approaches, fatigue life is characterized by crack initiation
and not specimen fracture.

SYSTEM DEFINATION
The definition of any system is an essential step in effectively modeling the system.
Extraneous information and components that may interfere with the evaluation must be
carefully screened. In addition, careful selection of the system is needed so that the
important elements of the system are not inadvertently omitted. The omission of vital
system components from the analysis could result in inaccurate or misleading findings. As
the preceding, example illustrates, the definition of system boundaries is an important first
step in performing risk assessment. The boundaries can be based on the objectives of the
analysis.

Generally, a marine equipment or ship or project can be modeled to include a segment of

its environment that interacts significantly with it to define an engineering system. The
boundaries of the system are drawn based on the goals and characteristics of the project,
the class of performances (including failures) under consideration, and the objectives of
the analysis. This primary step in assessing marine systems involves the definition of the
architecture of the system. The definition can be based on observations at different system
levels that are established based on the goals of the project. The observations can be
about the different elements (or components) of the system, interactions among these
PAGE 26
 elements, and the expected behavior of the system. Each level of knowledge that is
obtained about an engineering problem defines a system to represent the project. As
additional levels of knowledge are added to previous ones, higher epistemological levels of
system definition and description are possible which, taken together, form a hierarchy of
the system descriptions.

An epistemological hierarchy of systems requires a generalized treatment of uncertainty in

both the architecture of the system and the collected information. This treatment can be
based, in part, on probability and statistical concepts, as well as other related tools.
Therefore, engineering systems should be viewed with an understanding of the knowledge
content of collected information including its associated uncertainties. Also, the user
should understand the limitations of prediction models that result from inherent
insufficiency of models as a result of the assumptions used in developing them. The
uncertainty dimension in the analysis of engineering systems can result in valuable insight
or information that is necessary to make rational decisions. Such a decision process
considers the available information or knowledge, decision choices, alternative decision
outcomes, and associated uncertainties.

14) Hull girder and local stresses

The loading on ship hull girder and its structural components induce primary, secondary,
tertiary and local stresses. A ship hull girder among waves is subjected to:
i. Vertical shear force and bending moment,
ii. Horizontal shear force and bending moment
iii. Torsional moment
iv. Local loading
Hull girder shear force and
bending moment

Hull girder

Hull girder vertical shear

force and bending moment.
The primary and secondary
stresses acting on a ship
section and the double
bottom structure.

PAGE 27
Euler’s Beam Bending Theory is has been used by structural engineers in analyzing the
bending aspects of beams. Naval Architects have adopted this theory, but in a slightly
different way. Unlike civil structures, a ship structure (which will be referred to as hull
girder from now on) is always supported by an “elastic foundation” (sea surface). The
direction of buoyancy on the girder is upwards, and its longitudinal distribution depends
upon the longitudinal distribution of the ship’s underwater volume. Which means, there is
more buoyancy at the midship region than the fore and aft ends. This leads us to a
buoyancy distribution which looks similar to Figure below. There is another factor that
Buoyancy curve of a ship (Max surf )
contributes to the load on
the girder. It is the weight
that acts onto the hull
girder; the weight
comprising of individual
weights of hull steel,
machinery, outfit, cargo,
fuel oil, lube oil, fresh
water, ballast, and non-
fuel cargo. Depending
upon the longitudinal
distribution of these
weights and their
individual magnitudes,
we obtain the longitudinal
distribution of load on the girder, which is referred to as Weight Curve. It is this load
curve that holds utmost importance in the longitudinal strength aspect. The Load curve is
obtained by the difference between the buoyancy curve and the weight curve.

Load Curve
obtained from
buoyancy curve
and weight curve.
superimpose both
the graphs and
subtract the
magnitudes of
weight from the
buoyancy at every
point along the
length to obtain the
longitudinal
distribution of total
load on the girder

Notice how the direction of the net load may be upwards or downwards at different
positions along the hull girder length, depending upon the buoyancy and weight
distributions. It is from this stage of the analysis that Euler’s Beam Bending Theory comes
PAGE 28
of great use. The theory says, if we plot the magnitude of area under the load curve from
the aft end upto a certain point fore of the aft, we obtain the shear force acting on the hull
section at that point. A longitudinal plot of this parameter gives us the distribution of the
shear force, which is the SF Diagram of the ship at that loading condition. If such an area
integration is performed on the SF curve, we obtain the Bending Moment curve of the ship
for that loading condition.

SF and BM
Diagrams of
a ship

The important points and conclusions from the points in the diagrams are as follows: -

 Designers obtain the weight curve after developing the General Arrangement Plan of
the ship.
 The load curve is subject to change, depending on the various loading conditions of
the ship. For example, in fully loaded condition the load is generally more in the
parallel mid body of the ship, i.e the region where most of the cargo is stowed, be it
any type of a ship. But in ballast condition where aft and fore ballast tanks are to be
filled up, the weight curve changes its shape owing to the increased weight at the aft
and fore ends.
 Your loading manuals are basically a guide to load the ship by the design standards,
which are nothing but inferences to these different conditions of weight
distributions.
 The shear force on any transverse section of the hull girder is zero at the aft end,
fore end and midships. So, failure due to shearing is a least concern in these
regions.
 The bending moment is always maximum at the midships. It is due to this effect,
that the bending stress always reaches a maximum at the midship region of any
ship, irrespective of its loading condition. The magnitudes may vary, but this nature
is followed through any loading condition that the ship encounters in its lifetime.
 Owing to the maximum bending stresses occurring at the midships, designers
consider the bending moment of midships as a threshold for design with a certain
factor of safety.

PAGE 29
  If in any case of loading, the bending stress at any section of the hull exceeds the
bending strength of the material of the hull, it goes for a failure. So why midships? It
is because of the maximum bending moment always occurring at midships, that the
midship region is prone to exceed the threshold of bending strength of the material
in a given condition of improper loading.
 Grounding has many a times resulted in midship cracks or split offs. Why? Try
recalling what happens during grounding. Breaching occurs, resulting in unwanted
load distributions along the hull, which result in hogging or sagging, which are
nothing but modes of bending of the hull girder. So when you’ve seen ships split due
to grounding, it is the bending moment at the midships that had already exceeded
the strength of the hull material, and eventually led to the failure!

As with any cargo ship it is important to load the cargo so that stresses in the ship remain
at a minimum or at least evenly distributed. This is especially so with large bulk
carriers. All ships are designed with limitations imposed upon their operability to ensure
that the structural integrity is maintained. Therefore, exceeding these limitations may
result in over-stressing of the ship's structure which may lead to catastrophic failure.

The ship's approved loading manual provides a description of the operational loading
conditions upon which the design of the hull structure has been based. The loading
instrument provides a means to readily calculate the still water shear forces and bending
moments, in any load or ballast condition, and assess these values against the design
limits. A ship's structure is designed to withstand the static and dynamic loads likely to be
experienced by the ship throughout its service life. The loads acting on the hull structure
when a ship is floating in still (calm) water are static loads, one of the major ones being
created by the cargo.

The main hull stresses set up by the cargo are hogging, sagging and shearing. These can be
minimized by evenly distributing the cargo - homogenous loading.

Dynamic loads are those additional loads exerted on the ship's hull structure through the
action of the waves and the effects of the resultant ship motions (i.e. acceleration forces,
slamming and sloshing loads). Hogging and sagging forces are at a maximum when the
wave length is equal to the length of the ship. Fig: bulk carrier strain monitoring sensor

Sloshing loads may be induced on

the ship's internal structure through
the movement of the fluids in
tanks/holds whilst slamming of the
bottom shell structure forward may
occur due to emergence of the fore
end of the ship from the sea in heavy
weather.

Cargo over-loading in individual hold

spaces will increase the static stress
levels in the ship's structure and
reduce the strength capability of the
structure to sustain the dynamic
loads exerted in adverse sea
PAGE 30
conditions.

In harbour, where the ship is in sheltered water and is subjected to reduced dynamic loads,
the hull is permitted to carry a higher level of stress imposed by the static loads, so a
certain amount of difference in the loading of each hold is allowable.

Most modern bulkers have strain monitoring equipment so that hull stresses that cause
hull fractures as above are minimized.

Bending Moment
The bending moment is the amount of bending caused to the ship's hull by external forces.
For example, the bending moment is the highest in the midship section when the ship's
ends are supported by crests of a wave known as `sagging' or `positive bending'. When the
ship is riding the crest of a wave at its midships, the bending moment is known as
`hogging' or `negative bending'. Bending moments are measured in tonne- metres.

Shearing Force
When two external parallel forces act in opposite directions on any part of a structure to
break it apart or shear it, the forces are known as shearing forces and are measured in
tones. Shearing stress is, therefore, the stress that may break or shear the structure apart.

All classification society member bulk carriers are assigned with permissible still water
shear forces (SWSF) and still water bending moment (SWBM) limits. There are normally two
sets of permissible SWSF and SWBM limits assigned to each ship, namely:

1. Seagoing (at sea) SWSF and SWBM limits.

2. Harbour (in port) SWSF and SWBM limits.

The seagoing SWSF and SWBM limits are not to be exceeded when the ship puts to sea or
during any part of a seagoing voyage. In harbour, where the ship is in sheltered water and
is subjected to reduced dynamic loads, the hull girder is permitted to carry a higher level of
stress imposed by the static loads. The harbour SWSF and SWBM limits are not to be
exceeded during any stage of harbour cargo operations.

When a ship is floating in still water, the ship's lightweight (the weight of the ship's
structure and its machinery) and deadweight (all other weights, such as the weight of the
bunkers, ballast, provisions and cargo) are supported by the global buoyancy upthrust
acting on the exterior of the hull. Along the ship's length there will be local differences in
the vertical forces of buoyancy and the ship's weight. These unbalanced net vertical forces
acting along the length of the ship will cause the hull girder to shear and to bend, inducing
a vertical still water shear force (SWSF) and still water bending moment (SWBM) at each
section of the hull.

At sea, the ship is subjected to cyclical shearing and bending actions induced by
continuously changing wave pressures acting on the hull. These cyclical shearing and
PAGE 31
bending actions give rise to an additional component of dynamic, wave induced, shear force
and bending moment in the hull girder. At any one time, the hull girder is subjected to a
combination of still water and wave induced shear forces and bending moments.

The stresses in the hull section caused by these shearing forces and bending moments are
carried by continuous longitudinal structural members. These structural members are the
strength deck, side shell and bottom shell plating and longitudinals, inner bottom plating
and longitudinals, double bottom girders and topside and hopper tank sloping plating and
longitudinals, which are generally defined as the hull girder.

A hull girder in a seaway is subjected to a vertical shear force Fv and bending moment Mv
given by, Mv = Ms + Mw and Fv = Fs + Fw where

Fs and Ms = Stillwater shear force and bending moment

Fw and Mw = Wave shear force and bending moment

The hull girder bending stresses induced by a vertical bending moment in a sagging
condition is shown in fig
Hull girder stresses due to vertical bending
moment
The hull girder stress at the bottom and
deck plating are given by

Where

M = Total hull girder bending moment

I = Second moment of area of ship section
yB = Distance of bottom plating from ship
section neutral axis
yD = Distance of deck plating from ship section neutral axis.

The hull girder of a ship subjected to a horizontal bending moment MH will induce normal
stresses as shown in Fig. The horizontal bending moments induce tensile stresses in either
the port side shell and compressive stresses in the starboard side shell or vice versa.

PAGE 32
The hull girder stresses induced in the side shell plating is given by:
⋅
σB = (MH X B)/ Iy

Where:
MH = horizontal bending moment
B = ship breadth
Iy = second moment of area of ship section about the y-axis.

Unstable Fracture and Countermeasures

If subjected to high loads, structures like ships may essentially fail by two
mechanisms, plastic collapse governed by material yielding and buckling as well
as unstable fracture governed by material toughness. Actually, these two failure modes
interact. In this article, the main influencing factors on unstable fracture and measures for
its avoidance are outlined.

Types of Fracture and Influencing Factors

Generally, it may be distinguished between brittle fracture, showing a brittle surface and
negligible plastic deformation, and ductile fracture, being characterized by high plastic
deformation, showing an inclined fracture surface particularly in thin plates. Brittle
fracture is feared because the associated stress may be rather low. Well‐known factors
supporting particularly the occurrence of brittle fracture are:

 operation temperature
 stress level (due to external loads, residual, and temperature‐related stresses)
 stress state (multiaxiality and stress concentration)
 load and strain rate (less relevant in ships)
 material structure (also after heat treatment)

Stress concentration and multiaxiality are always present in welded ship structures,
particularly at weld imperfections and probable fatigue cracks. Therefore, unstable fracture
is a concern for designers as it tends to be catastrophic with severe consequences.
Fortunately, failures related to unstable fracture have been rarely observed during the past
years, also due to measures taken for their avoidance.

PAGE 33
Measures to Avoid Unstable Fracture
Most significant for the avoidance of unstable fracture is the material behavior, and in
particular, the toughness. This applies to base material as well as to weld material and the
adjacent heat‐affected zones (HAZ). Steel tends to become brittle at low temperatures. The
transition temperature between low and high toughness depends on the microstructure
which can be influenced by alloying elements and the steel‐making process. Therefore, steel
selection should be such that the transition temperature is well below the operation
temperature to ensure sufficiently high fracture toughness, at least in critical areas.

The mechanical properties of steel are characterized by its strength, for example, specified
yield strength of 235, 315, 355, or 395 MPa, and also by the steel grade within one strength
category, denoted A, B, D, E, and F (IACS, 2014). They differ mainly in the transition
temperature and, hence, in fracture toughness. This is checked by the so‐called notch
impact test (Charpy test), where small bars with notches (section 10 mm × 10 mm with 2
mm deep notch) are impacted at different temperatures, measuring the consumed impact
energy. Certain values (usually 27 J) are required for the different steel grades at given
temperatures (IACS, 2014). No requirement is specified for grade A.

The requirement of a necessary steel grade in a ship hull is regulated such that structural
members are categorized into secondary, primary, and special. The
category special requires the highest toughness and includes generally the longitudinal
structural members in the outer areas of the upper flange (deck stringer, sheer strake, and
longitudinal hatch coaming of container ships) and of the lower flange (bilge strake) of the
hull girder in addition to some other critical areas. Furthermore, the length‐wise location
(e.g., 0.4 L amidships) as well as the plate thickness determines the required material
grade. Thicker plates are known to be more critical with respect to brittle fracture
particularly as they support stress multiaxiality.

The high fracture toughness in special areas successfully helped to avoid unstable
fractures during the past years which had occurred in early times of welded ship
structures. Furthermore, they serve as crack arrestors for fractures occurring in other
areas with less toughness. High steel grades allow the application in special structures
down to temperatures of
about −55°C, as required
for example,
in LPG (liquified
petroleum gas) tankers.
Figure 9 shows an
example of required steel
grades in such a ship
type. Deeper
temperatures can be
accommodated only by
special steels or
aluminum alloys, which
are used in LNG (liquefied
natural gas) tankers at
temperatures of −163°C.

Figure 9 Steel grades selected for an LPG gas carrier. PAGE 34

Fracture Assessment Methods
The notch impact test mentioned above does not allow the quantitative assessment of the
risk of unstable fracture of flaws, defects, and fatigue cracks detected during fabrication or
operation. Instead, this requires a fitness‐for‐service assessment, more recently
named engineering critical assessment (ECA) based on fracture mechanics. This article gives
a short overview; more details can be found in codes and recommendations, for example,
BS 7910 2005 or Koçak 2007 and monographs, for example, MacDonald 2011.

Crack‐Driving Parameters
Flaws and defects occurring particularly in welds are usually regarded as cracks with sharp
tips, which is also typical for fatigue cracks. Here, the linear‐elastic stress field is singular
so that other parameters than stress are required.

The linear‐elastic fracture mechanics use the so‐called stress intensity factor as crack‐
driving parameter which is directly coupled with the energy release rate during crack
extension and with the path‐independent J‐integral around the crack tip, which can be
interpreted as the difference in potential energy of two identical bodies having cracks
slightly differing in length. All these parameters are suitable for the assessment of cracks
with small or negligible plastic zones which is usually the case during fatigue crack growth,
where the stress intensity factor can directly be used in the Paris–Erdogan law for cyclic
crack propagation.

However, unstable fracture of ductile materials is mostly coupled with significant plastic
deformations so that the critical stress intensity factor K IC is no more valid. More useful are
the critical J‐integral and the critical crack tip opening displacement (CTOD).

Owing to the energetic basis, the J‐integral is a suitable parameter to characterize the crack
resistance of the material also for elastic–plastic crack tip loading. The critical value J c is
determined with standardized specimens in fracture mechanics tests.

In the CTOD approach, the transverse extension δ at the crack tip is measured. Blunting
occurs here, where the material experiences large elastic–plastic strains until the critical
displacement δ c is reached and the unstable fracture with further crack propagation
occurs. CTOD has gained practical relevance, for example, in the design of offshore
structures where certain values are required for critical areas such as the heat‐affected
zone of welds in thick plates. Minimum values are typically a fraction of 1 mm.

Failure Assessment Diagram

In structures subjected to high loads, a general plastic failure can occur alternatively to or
in combination with fracture. As both failure modes interact with each other, the
assessment with regard to fracture is often performed today with an interaction curve
between them in the failure assessment diagram (FAD).

The vertical axis shows the crack‐driving parameter K r, that is, the ratio between the
linear‐elastic stress intensity factor K I and the crack resistance K mat of the material. The
latter can be the resistance against brittle fracture K Ic or the critical J‐integral or δ c, which
PAGE 35
is formally converted into K mat. Plotted on the horizontal axis is the plastification ratio L r,
which can be expressed by the external load F and the ultimate fully plastic load F u or
alternatively by corresponding stresses. The limit curve has been derived from extensive
finite element simulations and experiments, and has been implemented in the procedures
and recommendations mentioned (Figure 10).

From the FAD, a critical load or, else, crack

length can be derived allowing the risk of
unstable fracture to be assessed. Finally, it
should be noted that a correlation between the
crack resistance parameters and the transition
temperature for specific notch impact energies
is possible for ferritic and low‐alloy steels on
the basis of the master curve of the
relationship between crack resistance and
temperature by Wallin 1991

Figure 10 Failure assessment diagram (schematically).

PAGE 36